mirror of
https://github.com/crewAIInc/crewAI.git
synced 2026-01-10 00:28:31 +00:00
823f22a601
- Add environment variables for default embedding provider - Support Ollama as default embedding provider - Add tests for memory reset with different providers - Update documentation Fixes #2023 Co-Authored-By: Joe Moura <joao@crewai.com>
2.0 KiB
2.0 KiB
Memory Management in CrewAI
CrewAI provides a robust memory system that allows agents to store and retrieve information across conversations and tasks.
Memory Types
- Short Term Memory: Stores recent interactions and context
- Long Term Memory: Persists important information for extended periods
- Entity Memory: Tracks information about specific entities
Configuration
Embedding Providers
CrewAI supports multiple embedding providers for memory storage. By default, it uses OpenAI, but you can configure different providers:
# OpenAI (default)
export CREWAI_EMBEDDING_PROVIDER=openai
export CREWAI_EMBEDDING_MODEL=text-embedding-3-small
export OPENAI_API_KEY=your_key
# Ollama
export CREWAI_EMBEDDING_PROVIDER=ollama
export CREWAI_EMBEDDING_MODEL=llama2
export CREWAI_OLLAMA_URL=http://localhost:11434/api/embeddings # Optional
Memory Operations
Reset all memories:
crewai reset-memories -a
The memory system will use the configured embedding provider for all operations, including memory reset.
Troubleshooting
Common Issues
- OpenAI API Key Missing
# Error: EmbeddingConfigurationError: Invalid configuration for OpenAI provider
# Solution: Set your OpenAI API key
export OPENAI_API_KEY=your_key
- Ollama Connection Issues
# Error: Failed to connect to Ollama server
# Solution: Ensure Ollama is running and accessible
curl http://localhost:11434/api/embeddings
- Memory Reset Permission Issues
# Error: Failed to reset memory (readonly database)
# Solution: Check file permissions or use custom path
export CREWAI_MEMORY_PATH=/path/with/write/access
Custom Storage Path
You can configure a custom storage path for memory files:
from crewai.memory import ShortTermMemory
memory = ShortTermMemory(path="/custom/storage/path")
This is useful when:
- Default storage location has permission issues
- You want to isolate memory storage for different crews
- You need to manage memory persistence manually